2025 Volume 81 Issue 21 Article ID: 25-21005
Maintenance of expressway pavements in Japan tends to become more frequent due to the occurrence of fatigue cracks originating from the bottom of the base course caused by aging, as well as the deepening of damage resulting from water infiltration through surface cracks. To implement preventive maintenance before such damage progresses, it is important to evaluate the bearing capacity of the asphalt layers.
However, assessment using Falling Weight Deflectometer (FWD) measurements requires traffic restrictions, making it difficult to perform long-distance evaluations in a short time, as is possible with road surface condition survey vehicles. In this study, we improved a deflection index estimation model based on machine learning using data from the Pavement Management System (PMS), aiming to enhance its versatility and prediction accuracy.
As a result, the developed model achieved a coefficient of determination (R²) of 0.794 between actual and estimated values, indicating the potential to rapidly assess the sublayer bearing capacity over long sections, such as approximately 250 km of the Hokuriku Expressway.
